Prostate cancer (PCa) is the most common cancer diagnosed in a diverse population of men, and patient outcomes are heterogeneous. Androgen signaling inhibitors are the frontline treatment for advanced PCa. However, the duration of benefit from this treatment is widely variable, and eventually all PCa develops resistance. Survival for men with metastatic castrate resistant PCa remains poor with limited treatment options. Thus, it is imperative to develop novel biomarkers to identify men at greatest risk of developing lethal disease early in their diagnosis to better tailor treatments. Alterations in cell metabolism are key drivers of PCa aggressiveness and resistance to anti-androgen therapy....
Read More
Prostate cancer (PCa) is the most common cancer diagnosed in a diverse population of men, and patient outcomes are heterogeneous. Androgen signaling inhibitors are the frontline treatment for advanced PCa. However, the duration of benefit from this treatment is widely variable, and eventually all PCa develops resistance. Survival for men with metastatic castrate resistant PCa remains poor with limited treatment options. Thus, it is imperative to develop novel biomarkers to identify men at greatest risk of developing lethal disease early in their diagnosis to better tailor treatments. Alterations in cell metabolism are key drivers of PCa aggressiveness and resistance to anti-androgen therapy. Biomarkers that reflect dynamic metabolic dysfunction in the treatment setting hold the potential to better stratify patients, detect early resistance to therapy, and adjust course as needed. Detection of metabolic changes via advanced imaging techniques holds great promise to monitor metabolic processes associated with therapy resistance. The resulting multi-modal data is powerful in deconvoluting heterogeneity in patient responses to treatment and the underlying mechanisms across the disease spectrum. Novel computational models are needed to harness this data and categorize PCa into biologically-informed sub-types. However, this large-scale data where each measurement may be characterized by hundreds of variables poses challenges for classification methods, and traditional statistical and engineering approaches often fail in this setting. We plan to apply artificial intelligence, specifically neural machine learning in this context of anatomical, metabolic and multiplatform omics to identify different sub-types of lethal PCa. The overarching goal is to apply neural map-based machine learning to achieve accurate classification of PCa by combining preclinical and clinical data to develop non-invasive predictive biomarkers and bring precision medicine from promise to reality.
Read Less
